1. 说明

本文基于：spark-2.4.0-hadoop2.7-高可用(HA)安装部署

2. 启动Spark Shell

　　在任意一台有spark的机器上执行

 1 # --master spark://mini02:7077  连接spark的master，这个master的状态为alive，而不是standby
 2 # --total-executor-cores 2  总共占用2核CPU
 3 # --executor-memory 512m  每个woker占用512m内存
 4 [yun@mini03 ~]$ spark-shell --master spark://mini02:7077 --total-executor-cores 2 --executor-memory 512m  
 5 2018-11-25 12:07:39 WARN  NativeCodeLoader:62 - Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
 6 Setting default log level to "WARN".
 7 To adjust logging level use sc.setLogLevel(newLevel). For SparkR, use setLogLevel(newLevel).
 8 Spark context Web UI available at http://mini03:4040
 9 Spark context available as 'sc' (master = spark://mini02:7077, app id = app-20181125120746-0001).
10 Spark session available as 'spark'.
11 Welcome to
12       ____              __
13      / __/__  ___ _____/ /__
14     _\ \/ _ \/ _ `/ __/  '_/
15    /___/ .__/\_,_/_/ /_/\_\   version 2.4.0
16       /_/
17          
18 Using Scala version 2.11.12 (Java HotSpot(TM) 64-Bit Server VM, Java 1.8.0_112)
19 Type in expressions to have them evaluated.
20 Type :help for more information.
21 
22 scala> sc
23 res0: org.apache.spark.SparkContext = org.apache.spark.SparkContext@77e1b84c

注意：

　　如果启动spark shell时没有指定master地址，但是也可以正常启动spark shell和执行spark shell中的程序，其实是启动了spark的local模式，该模式仅在本机启动一个进程，没有与集群建立联系。

2.1. 相关截图

3. 执行第一个spark程序

　　该算法是利用蒙特?卡罗算法求PI

 1 [yun@mini03 ~]$ spark-submit  2  --class org.apache.spark.examples.SparkPi  3  --master spark://mini02:7077 \
 4  --total-executor-cores 2  5  --executor-memory 512m  6  /app/spark/examples/jars/spark-examples_2.11-2.4.0.jar 100
 7 # 打印的信息如下：
 8 2018-11-25 12:25:42 WARN  NativeCodeLoader:62 - Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
 9 2018-11-25 12:25:43 INFO  SparkContext:54 - Running Spark version 2.4.0
10 ………………
11 2018-11-25 12:25:49 INFO  TaskSetManager:54 - Finished task 97.0 in stage 0.0 (TID 97) in 20 ms on 172.16.1.14 (executor 0) (98/100)
12 2018-11-25 12:25:49 INFO  TaskSetManager:54 - Finished task 98.0 in stage 0.0 (TID 98) in 26 ms on 172.16.1.13 (executor 1) (99/100)
13 2018-11-25 12:25:49 INFO  TaskSetManager:54 - Finished task 99.0 in stage 0.0 (TID 99) in 25 ms on 172.16.1.14 (executor 0) (100/100)
14 2018-11-25 12:25:49 INFO  TaskSchedulerImpl:54 - Removed TaskSet 0.0, whose tasks have all completed, from pool 
15 2018-11-25 12:25:49 INFO  DAGScheduler:54 - ResultStage 0 (reduce at SparkPi.scala:38) finished in 3.881 s
16 2018-11-25 12:25:49 INFO  DAGScheduler:54 - Job 0 finished: reduce at SparkPi.scala:38, took 4.042591 s
17 Pi is roughly 3.1412699141269913
18 ………………

4. Spark shell求Word count 【结合Hadoop】

1、启动Hadoop

2、将文件放到Hadoop中

 1 [yun@mini05 sparkwordcount]$ cat wc.info 
 2 zhang linux
 3 linux tom
 4 zhan kitty
 5 tom  linux
 6 [yun@mini05 sparkwordcount]$ hdfs dfs -ls /
 7 Found 4 items
 8 drwxr-xr-x   - yun supergroup          0 2018-11-16 11:36 /hbase
 9 drwx------   - yun supergroup          0 2018-11-14 23:42 /tmp
10 drwxr-xr-x   - yun supergroup          0 2018-11-14 23:42 /wordcount
11 -rw-r--r--   3 yun supergroup   16402010 2018-11-14 23:39 /zookeeper-3.4.5.tar.gz
12 [yun@mini05 sparkwordcount]$ hdfs dfs -mkdir -p /sparkwordcount/input
13 [yun@mini05 sparkwordcount]$ hdfs dfs -put wc.info /sparkwordcount/input/1.info
14 [yun@mini05 sparkwordcount]$ hdfs dfs -put wc.info /sparkwordcount/input/2.info
15 [yun@mini05 sparkwordcount]$ hdfs dfs -put wc.info /sparkwordcount/input/3.info
16 [yun@mini05 sparkwordcount]$ hdfs dfs -put wc.info /sparkwordcount/input/4.info
17 [yun@mini05 sparkwordcount]$ hdfs dfs -ls /sparkwordcount/input
18 Found 4 items
19 -rw-r--r--   3 yun supergroup         45 2018-11-25 14:41 /sparkwordcount/input/1.info
20 -rw-r--r--   3 yun supergroup         45 2018-11-25 14:41 /sparkwordcount/input/2.info
21 -rw-r--r--   3 yun supergroup         45 2018-11-25 14:41 /sparkwordcount/input/3.info
22 -rw-r--r--   3 yun supergroup         45 2018-11-25 14:41 /sparkwordcount/input/4.info

3、进入spark shell命令行，并计算

1 [yun@mini03 ~]$ spark-shell --master spark://mini02:7077 --total-executor-cores 2 --executor-memory 512m  
2 # 计算完毕后，打印在命令行
3 scala> sc.textFile("hdfs://mini01:9000/sparkwordcount/input").flatMap(_.split(" ")).map((_, 1)).reduceByKey(_+_).sortBy(_._2, false).collect  
4 res6: Array[(String, Int)] = Array((linux,12), (tom,8), (kitty,4), (zhan,4), ("",4), (zhang,4))  
5 # 计算完毕后，保存在HDFS【因为有多个文件组成，则有多个reduce，所以输出有多个文件】
6 scala> sc.textFile("hdfs://mini01:9000/sparkwordcount/input").flatMap(_.split(" ")).map((_, 1)).reduceByKey(_+_).sortBy(_._2, false).saveAsTextFile("hdfs://mini01:9000/sparkwordcount/output")
7 # 计算完毕后，保存在HDFS【将reduce设置为1，输出就只有一个文件】
8 scala> sc.textFile("hdfs://mini01:9000/sparkwordcount/input").flatMap(_.split(" ")).map((_, 1)).reduceByKey(_+_, 1).sortBy(_._2, false).saveAsTextFile("hdfs://mini01:9000/sparkwordcount/output1")

4、在HDFS的查看结算结果

 1 [yun@mini05 sparkwordcount]$ hdfs dfs -ls /sparkwordcount/
 2 Found 2 items
 3 drwxr-xr-x   - yun supergroup          0 2018-11-25 15:03 /sparkwordcount/input
 4 drwxr-xr-x   - yun supergroup          0 2018-11-25 15:05 /sparkwordcount/output
 5 drwxr-xr-x   - yun supergroup          0 2018-11-25 15:07 /sparkwordcount/output1
 6 [yun@mini05 sparkwordcount]$ hdfs dfs -ls /sparkwordcount/output
 7 Found 5 items
 8 -rw-r--r--   3 yun supergroup          0 2018-11-25 15:05 /sparkwordcount/output/_SUCCESS
 9 -rw-r--r--   3 yun supergroup          0 2018-11-25 15:05 /sparkwordcount/output/part-00000
10 -rw-r--r--   3 yun supergroup         11 2018-11-25 15:05 /sparkwordcount/output/part-00001
11 -rw-r--r--   3 yun supergroup          8 2018-11-25 15:05 /sparkwordcount/output/part-00002
12 -rw-r--r--   3 yun supergroup         34 2018-11-25 15:05 /sparkwordcount/output/part-00003
13 [yun@mini05 sparkwordcount]$ 
14 [yun@mini05 sparkwordcount]$ hdfs dfs -cat /sparkwordcount/output/part*
15 (linux,12)
16 (tom,8)
17 (,4)
18 (zhang,4)
19 (kitty,4)
20 (zhan,4)
21 ###############################################
22 [yun@mini05 sparkwordcount]$ hdfs dfs -ls /sparkwordcount/output1
23 Found 2 items
24 -rw-r--r--   3 yun supergroup          0 2018-11-25 15:07 /sparkwordcount/output1/_SUCCESS
25 -rw-r--r--   3 yun supergroup         53 2018-11-25 15:07 /sparkwordcount/output1/part-00000
26 [yun@mini05 sparkwordcount]$ hdfs dfs -cat /sparkwordcount/output1/part-00000 
27 (linux,12)
28 (tom,8)
29 (,4)
30 (zhang,4)
31 (kitty,4)
32 (zhan,4)